field note grand rapids, mi · 2026
ai program lead · behavior change · learning systems

ai rollouts
that humans
actually adopt.

Most AI rollouts change nothing. I lead the kind that do. I'm Emily Green. 15 years of behavior change and learning systems for Fortune 500s. Now leading AI programs the same way. The whole caboodle.

see my work
trusted by fortune 500 8+ enterprise teams
selected work

engagements.
filed and catalogued.

Programs and learning systems that drive measurable behavior change, and the artifacts to prove it.

ENG-INTUIT 10,000+ employees
intuit · career growth ai recommendations + structured learning paths

ai-powered career growth for 10k+ employees.

A comprehensive career growth program integrating AI recommendations with structured learning paths, helping employees navigate their professional development journey.

open case file →
10K+ employees · ai-powered career growth
caboodledesign.info/work/intuit.html
ENG-APPLE 251 companies · 31 countries
apple · supplier code of conduct ethical, environmental, innovation standards

supplier training across 251 companies, 31 countries.

A global training program for Apple's supplier network covering ethical standards, environmental compliance, and innovation requirements. Built to translate, measure, and scale.

open case file →
95% completion across 251 suppliers
caboodledesign.info/work/apple.html
PROJ-PP001 self-initiated · 2026
the people's professors · capstone research graphic design · custom code · tiktok

a thesis about tiktok. so i built the proof.

A capstone project using graphic design and custom code to deliver a one-of-a-kind learning experience. Three conference keynotes followed.

open case file →
3× conference keynotes · 2026
caboodledesign.info/work/peoples-professors.html
ENG-GITLAB 2 months · nl · de · fr
gitlab · works council compliance interactive decision tools · lms-embedded html

works council training delivered in 2 months.

Interactive decision tools and contact guides for Netherlands, Germany, and France. HTML-based learning components embedded in the LMS with localized content for each region.

open case file →
60d kickoff to launch · nl · de · fr
caboodledesign.info/work/gitlab.html
ENG-TMOBILE corporate legal
t-mobile · legal affairs branching simulation · regulatory clarity

legal training that cut inquiries 40%.

Interactive legal compliance training for T-Mobile's corporate teams. Regulatory requirements translated into clear, actionable scenarios practiced as branching simulations.

open case file →
40% reduction in legal inquiries post-rollout
caboodledesign.info/work/tmobile.html
ENG-TRUST20 5 weeks · multi-state
trust20 · gordon food services scenario-based · multi-state accreditation

accredited alcohol server training in 5 weeks.

Scalable, scenario-based alcohol server training meeting multi-state accreditation requirements for Gordon Food Services within a five-week timeline.

open case file →
5+ states, multi-region accreditation rollout
caboodledesign.info/work/trust20.html

See a problem here that looks like yours? Let's talk.

approach

the whole caboodle.

Where strategy, media, and behavioral science converge. Three methods, applied in order, borrowed from anthropology, behavioral science, and editorial media. Not from the LMS vendor catalog.

ethnographic research

I observe what people actually do before I design anything.

Before a single storyboard, I study what people actually do in their workplace context. This anthropological approach ensures training addresses real behavior, not assumptions. I watch. I listen. I ask the uncomfortable questions.

  • contextual inquiry · stakeholder interviews · artifact analysis
  • global & compliance: localization, wcag, regulatory standards
  • curriculum architecture: outcome-driven learning paths
behavioral strategy

I design for behavior change. Then I measure it.

Action Mapping for design. A layered Kirkpatrick-Phillips evaluation model for measurement, from learner reaction all the way to demonstrated ROI for the business. Nudges, motivation design, habit loops, AI integration. The thinking is the job.

  • action mapping · kirkpatrick-phillips · choice architecture
  • measurement instrumented before launch, not after
  • ai as a field tool, never as a substitution
creative media

Scalable learning systems, not one-off decks.

Interaction design, video, motion graphics, visual storytelling. Design systems and media libraries and interactive experiences. Every asset is built to evolve with your organization, so the training investment future-proofs itself.

  • remotion · three.js · storyline · rise · editorial web
  • single source of truth asset systems
  • accessibility and retention measured, not assumed
design meets measurement

design with action mapping.
measure with kirkpatrick.

Every engagement ladders up through the same four rungs. The business goal lines up with the Kirkpatrick level, and a real client proves the row.

01

business goal

What outcome must change?

L4

results

Did it matter?

02

critical actions

What must people do differently?

L3

behavior

Are they doing it?

03

practice activities

How do they rehearse it?

L2

learning

Did they learn it?

04

essential info

What's the minimum to act?

L1

reaction

Did they like it?

completion is the metric everyone reports; behavior is the one that matters.

caboodle labs

where behavior design meets ai experimentation.

Personal projects, prototypes, and proofs-of-concept. Each one exploring whether AI can solve a real L&D workflow problem, not just build features.

hover to preview click to open

ai prototype · applied llm + rag

greenroom: a development copilot

A development copilot built around behavior change, not just content. Real RAG with citations, an agentic routing loop, and a 22-case eval harness. Scoped, built, and evaluated solo.

claude code · rag · eval harness view experiment →
ai prototype · workflow

inline stakeholder review system

A feedback workspace built directly into a client landing page. Pins, color-coded reviewers, a slide-out panel grouping comments by section with resolve tracking. No email threads. No annotated screenshots.

claude code · vibe coding view experiment →
ai prototype · feedback design

emotion-aware feedback system

Exploring how learner emotional state can inform the type, tone, and timing of feedback. Self-reported check-ins plus adaptive response patterns, so feedback lands, not just lands accurately.

adaptive ux · emotion detection view experiment →
ai prototype · systems thinking

content ops dashboard

An entire content operations stack automated with Claude Code. Daily briefs, Canva carousel generation, research pipelines, analytics reports. Every piece maps to a problem learning designers face daily.

claude code · mcp · canva view experiment →
ai prototype · programmatic media

programmatic video & content automation

Remotion templates plus MCP-driven Canva generation. Template to published content with no manual design work. Testing whether programmatic media can replace one-off production for L&D teams.

remotion · mcp · canva view experiment →
ai prototype · microlearning

context-aware microlearning

Pick a work scenario, receive a contextually relevant micro-intervention. The right nudge at the right moment, not just shorter content. The 2026 evolution of microlearning.

moment-of-need · behavioral nudges view experiment →
stakeholder review system labs/stakeholder-review.html
Emily Green, founder of caboodle design
subject · emily green · founder
about emily

15 years designing for behavior change.
one stubborn conviction: ai programs fail without it.

I started caboodle design because I was tired of seeing organizations pour resources into programs that checked a box but never changed a behavior. The same trap is happening with AI rollouts right now. My work sits at the intersection of behavioral science, AI program leadership, and learning systems. It's the whole caboodle.

years designing for behavior change
15+
fortune 500 clients
8+
avg completion
95%
countries reached
31+
more about emily · speaking, full bio, capabilities
next field note

ready to create
learning that actually
changes behavior?

Send the shape of the problem. I'll send back whether training is actually the answer, and what might work better if it isn't.

start a conversation